为什么我们要思考：测试时计算与思维链的进展

AI 摘要

Lil'Log 发布文章《Why We Think》，回顾了测试时计算和思维链（CoT）在提升模型性能方面的最新进展。文章从人类双过程理论出发，将慢思考类比为测试时计算，并探讨了通过并行采样和顺序修正等解码策略来优化模型输出的方法。文章还提到了强化学习在提升 CoT 推理能力中的应用，并引用了 o1、o3 和 R1 等模型的技术报告。

正文节选

Special thanks to John Schulman for a lot of super valuable feedback and direct edits on this post. Test time compute (Graves et al. 2016, Ling, et al. 2017, Cobbe et al. 2021) and Chain-of-thought (CoT) (Wei et al. 2022, Nye et al. 2021), have led to significant improvements in model performance, while raising many research questions. This post aims to review recent developments in how to effectively use test-time compute (i.e. “thinking time”) and why it helps. The core idea is deeply connecte

发布时间：2025-05-01 08:00

抓取时间：2026-08-02 00:26

来源机构：Lilian Weng